Drug candidates are often identified from natural sources or combinatorial libraries. NMR is the primary tool for their structure elucidation. Improving sensitivity and automating the structure elucidation are important to accelerate the drug discovery and development process. This project uses inverse-detected heteronuclear NMR correlation experiments. This increases the carbon skeleton detection sensitivity forty times compared to traditional 2D INADEQUATE spectra and thereby saves three orders of magnitude of acquisition time. Our NMRanalyst software automates the analysis of NMR spectra with a reliability and sensitivity often exceeding its visual analysis. A bond between NMR active heteroatoms (e.g., carbon, nitrogen) is observable when one bonded atom is protonated. Unobserved bonds can still be located based on observed longer-range heteronuclear NMR correlations. Our AssembleIt module derives the most likely carbon skeleton(s) from the analysis results despite some unobserved correlations or inconsistencies between observed ones. Oxygen, sulfur, and most halogens are not NMR observable. But they influence the shift of surrounding observable atoms. Using chemical shift prediction, this project aims to extend determined carbon skeletons to complete molecular structures including heteroatoms and bond multiplicities. It further aims to automate molecular structure determinations. Molecular features insufficiently characterized by the input data can be reported for further human inspection. [unreadable] [unreadable]